This is a Case Study carried on a dataset on Udemy Courses. I am going to Analyse Data Collected on Courses uploaded on UDEMY. It will use the six steps in Data Analysis. ASK, PREPARE, PROCESS, ANALYZE, SHARE, and ACT

1. Ask Phase

The client is a fictional company called SubAdd Tech Solutions, which seeks to analyse existing trends and various variables affecting the performance of courses on UDEMY to help them strategize for the courses they are about to create. They have no Specific Questions, they want a thorough investigation of the complete dataset as much as is possible with a Junior Data Analyst.

Secondary Data will be used since SubAdd tech Solutions do not have any form of pre-existing data already.

1.1 Stakeholders

  • Joseph Salmawobil: Content creation manager at SubAdd Tech Solutions, he is interested in knowing what areas will be relevant to create courses around that will be more profitable and How long courses should be.
  • Joe Yapzor: The Marketing Director, who wants to know how to strategically use discounting and various Pricing strategies to sell the course. He is also interested in knowing how reviews can affect the success of a course.
  • Josiah Wunmoya: Chief Data Analyst, who was supposed to do this analysis but is currently busily working on a data set on marketing of the newest product launched by SubAdd Tech Solution, to help scale it up. He will however review the work at the end of it to give feedback.

1.2 Deliverables

  • A Markdown version of the analysis process for the Chief Data Analyst to review.
  • A PowerPoint Presentation to help the Content Creation and Marketing Director to have get insights from the Analysis

2. Prepare Phase

Since the company does not have already existing data for the analysis, a search was carried out online and a dataset was obtained from Kaggle. Kaggle is a community of Data Scientist, which contains a lot of Datasets to work with, some of which are free.

2.1 Data Source

Secondary data was collected from a publicly available dataset on kaggle named Business Courses Udemy. It Contains information on 10,000 Courses on Udemy. All courses in the data set are paid. This Data Set was Downloaded freely as a csv file with the name udemy_output_All_Business_p1_p626 The Dataset contains 20 columns consisting of different Variables which describe the data. The number of roles are 9447, excluding the first column which serves as the header. It includes various metrics such as discounts, course published dates, course prices, number of reviews, rating, and number of subscribers among others.

3. Process Phase

3.0 Tool for Analysis

R Programming Language will be used for this analysis, this is because of R’s ability to

  • Process Large Datasets faster.
  • Allow for reproducibility of analysis by rerunning the code.
  • Allow for manipulation of Data in various ways.
  • Ability to create world class Data Visualizations among many other advantages.

Python can also be used for this analysis, but R is the official language of the company. R Studio desktop version is used for this analysis.

3.1 Setting Up the Environment for Analysis

#Installing Packages 
# Remove the # and Run these lines if you have not used these packages before

#install.packages("tidyverse")
#install.packages("lubridate")
#install.packages("readr")


# Loading of Packages
library(tidyverse)
## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr     1.1.2     ✔ readr     2.1.4
## ✔ forcats   1.0.0     ✔ stringr   1.5.0
## ✔ ggplot2   3.4.2     ✔ tibble    3.2.1
## ✔ lubridate 1.9.2     ✔ tidyr     1.3.0
## ✔ purrr     1.0.1     
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag()    masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
library(lubridate)
library(readr)
library(dplyr)

# tidyverse contain most of the tools that will be used in this analysis

3.2 Loading Dataset into the Envronment

# The Dataset is in the same folder as this MarkDown Book

courses <- read_csv("udemy_output_All_Business_p1_p626.csv")
## Rows: 9447 Columns: 20
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr   (6): title, url, discount_price__currency, discount_price__price_strin...
## dbl  (10): id, num_subscribers, avg_rating, avg_rating_recent, rating, num_r...
## lgl   (2): is_paid, is_wishlisted
## dttm  (2): created, published_time
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

Previewing the Data

head(courses)
glimpse(courses)
## Rows: 9,447
## Columns: 20
## $ id                           <dbl> 762616, 937678, 1361790, 648826, 637930, …
## $ title                        <chr> "The Complete SQL Bootcamp 2020: Go from …
## $ url                          <chr> "/course/the-complete-sql-bootcamp/", "/c…
## $ is_paid                      <lgl> TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,…
## $ num_subscribers              <dbl> 295509, 209070, 155282, 245860, 374836, 1…
## $ avg_rating                   <dbl> 4.66019, 4.58956, 4.59491, 4.54407, 4.470…
## $ avg_rating_recent            <dbl> 4.67874, 4.60015, 4.59326, 4.53772, 4.471…
## $ rating                       <dbl> 4.67874, 4.60015, 4.59326, 4.53772, 4.471…
## $ num_reviews                  <dbl> 78006, 54581, 52653, 46447, 41630, 38093,…
## $ is_wishlisted                <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,…
## $ num_published_lectures       <dbl> 84, 78, 292, 338, 83, 275, 23, 275, 144, …
## $ num_published_practice_tests <dbl> 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…
## $ created                      <dttm> 2016-02-14 22:57:48, 2016-08-22 12:10:18…
## $ published_time               <dttm> 2016-04-06 05:16:11, 2016-08-23 16:59:49…
## $ discount_price__amount       <dbl> 455, 455, 455, 455, 455, 455, 455, 455, 4…
## $ discount_price__currency     <chr> "INR", "INR", "INR", "INR", "INR", "INR",…
## $ discount_price__price_string <chr> "₹455", "₹455", "₹455", "₹455", "₹455", "…
## $ price_detail__amount         <dbl> 8640, 8640, 8640, 8640, 8640, 8640, 8640,…
## $ price_detail__currency       <chr> "INR", "INR", "INR", "INR", "INR", "INR",…
## $ price_detail__price_string   <chr> "₹8,640", "₹8,640", "₹8,640", "₹8,640", "…

3.3 Data Cleaning

This process is used to ensure data integrity and to avoid errors and biasis in the analysis

3.3.1 Checking for NA
#This checks for Null Values in the dataset
nulls = is.na(courses)

#This summarises the characteristics of the null values
summary(nulls)
##      id            title            url           is_paid       
##  Mode :logical   Mode :logical   Mode :logical   Mode :logical  
##  FALSE:9447      FALSE:9447      FALSE:9447      FALSE:9447     
##                                                                 
##  num_subscribers avg_rating      avg_rating_recent   rating       
##  Mode :logical   Mode :logical   Mode :logical     Mode :logical  
##  FALSE:9447      FALSE:9447      FALSE:9447        FALSE:9447     
##                                                                   
##  num_reviews     is_wishlisted   num_published_lectures
##  Mode :logical   Mode :logical   Mode :logical         
##  FALSE:9447      FALSE:9447      FALSE:9447            
##                                                        
##  num_published_practice_tests  created        published_time 
##  Mode :logical                Mode :logical   Mode :logical  
##  FALSE:9447                   FALSE:9447      FALSE:9447     
##                                                              
##  discount_price__amount discount_price__currency discount_price__price_string
##  Mode :logical          Mode :logical            Mode :logical               
##  FALSE:8937             FALSE:8937               FALSE:8937                  
##  TRUE :510              TRUE :510                TRUE :510                   
##  price_detail__amount price_detail__currency price_detail__price_string
##  Mode :logical        Mode :logical          Mode :logical             
##  FALSE:9447           FALSE:9447             FALSE:9447                
##
  • From the Results, NULL values exist only in the section that has to do with Discounts, which is ok because not all the courses have discount.
  • The Null values in the data does not comprisise it’s integrity, it rather confirms it
3.3.2 Checking for duplicates

This is to check is there are repeated rows in the dataset

#This checks for duplicated rows in the dataset
nulls = duplicated(courses)

#This summarises the characteristics of the duplicated rows
summary(nulls)
##    Mode   FALSE 
## logical    9447
  • No Duplicate case is found in the dataset.
3.3.3 Checking Columns to see if they are rightly categorised
glimpse(courses)
## Rows: 9,447
## Columns: 20
## $ id                           <dbl> 762616, 937678, 1361790, 648826, 637930, …
## $ title                        <chr> "The Complete SQL Bootcamp 2020: Go from …
## $ url                          <chr> "/course/the-complete-sql-bootcamp/", "/c…
## $ is_paid                      <lgl> TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,…
## $ num_subscribers              <dbl> 295509, 209070, 155282, 245860, 374836, 1…
## $ avg_rating                   <dbl> 4.66019, 4.58956, 4.59491, 4.54407, 4.470…
## $ avg_rating_recent            <dbl> 4.67874, 4.60015, 4.59326, 4.53772, 4.471…
## $ rating                       <dbl> 4.67874, 4.60015, 4.59326, 4.53772, 4.471…
## $ num_reviews                  <dbl> 78006, 54581, 52653, 46447, 41630, 38093,…
## $ is_wishlisted                <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,…
## $ num_published_lectures       <dbl> 84, 78, 292, 338, 83, 275, 23, 275, 144, …
## $ num_published_practice_tests <dbl> 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…
## $ created                      <dttm> 2016-02-14 22:57:48, 2016-08-22 12:10:18…
## $ published_time               <dttm> 2016-04-06 05:16:11, 2016-08-23 16:59:49…
## $ discount_price__amount       <dbl> 455, 455, 455, 455, 455, 455, 455, 455, 4…
## $ discount_price__currency     <chr> "INR", "INR", "INR", "INR", "INR", "INR",…
## $ discount_price__price_string <chr> "₹455", "₹455", "₹455", "₹455", "₹455", "…
## $ price_detail__amount         <dbl> 8640, 8640, 8640, 8640, 8640, 8640, 8640,…
## $ price_detail__currency       <chr> "INR", "INR", "INR", "INR", "INR", "INR",…
## $ price_detail__price_string   <chr> "₹8,640", "₹8,640", "₹8,640", "₹8,640", "…

  • All columns are in the right data class, including the published time and date created, which are in DateTime format

  • There is data integrity, hence I will proceed to the ANALYSIS phase.

4. Analyse Phase

Below are the analysis carried out on the dataset to get insights.

4.1 Descriptive Statistics

Descriptive Statistics gives an overview of various columns in the dataset.

# The Summary Function gives an overview statistics of all the non-chracter coluumns
summary(courses)
##        id             title               url            is_paid       
##  Min.   :   2762   Length:9447        Length:9447        Mode:logical  
##  1st Qu.: 673031   Class :character   Class :character   TRUE:9447     
##  Median :1381806   Mode  :character   Mode  :character                 
##  Mean   :1456603                                                       
##  3rd Qu.:2154793                                                       
##  Max.   :3469452                                                       
##                                                                        
##  num_subscribers    avg_rating    avg_rating_recent     rating     
##  Min.   :     3   Min.   :1.000   Min.   :1.000     Min.   :1.000  
##  1st Qu.:   141   1st Qu.:3.900   1st Qu.:3.879     1st Qu.:3.879  
##  Median :   916   Median :4.200   Median :4.206     Median :4.206  
##  Mean   :  3399   Mean   :4.139   Mean   :4.125     Mean   :4.125  
##  3rd Qu.:  2890   3rd Qu.:4.450   3rd Qu.:4.447     3rd Qu.:4.447  
##  Max.   :374836   Max.   :5.000   Max.   :5.000     Max.   :5.000  
##                                                                    
##   num_reviews      is_wishlisted   num_published_lectures
##  Min.   :    3.0   Mode :logical   Min.   :  0.0         
##  1st Qu.:   12.0   FALSE:9447      1st Qu.: 13.0         
##  Median :   33.0                   Median : 23.0         
##  Mean   :  299.1                   Mean   : 34.4         
##  3rd Qu.:  110.0                   3rd Qu.: 39.0         
##  Max.   :78006.0                   Max.   :699.0         
##                                                          
##  num_published_practice_tests    created                      
##  Min.   :0.0000               Min.   :2010-04-14 21:32:46.00  
##  1st Qu.:0.0000               1st Qu.:2015-11-18 07:28:59.50  
##  Median :0.0000               Median :2017-10-05 17:06:12.00  
##  Mean   :0.1018               Mean   :2017-06-24 23:21:45.66  
##  3rd Qu.:0.0000               3rd Qu.:2019-01-16 11:34:46.00  
##  Max.   :6.0000               Max.   :2020-09-02 13:25:24.00  
##                                                               
##  published_time                   discount_price__amount
##  Min.   :2010-04-14 16:32:46.00   Min.   : 455          
##  1st Qu.:2016-01-25 07:43:43.00   1st Qu.: 455          
##  Median :2017-11-30 18:57:20.00   Median : 455          
##  Mean   :2017-08-17 22:57:47.16   Mean   : 489          
##  3rd Qu.:2019-03-08 17:51:10.50   3rd Qu.: 455          
##  Max.   :2020-09-08 21:04:10.00   Max.   :3200          
##                                   NA's   :510           
##  discount_price__currency discount_price__price_string price_detail__amount
##  Length:9447              Length:9447                  Min.   : 1280       
##  Class :character         Class :character             1st Qu.: 1600       
##  Mode  :character         Mode  :character             Median : 3840       
##                                                        Mean   : 4829       
##                                                        3rd Qu.: 8640       
##                                                        Max.   :12800       
##                                                                            
##  price_detail__currency price_detail__price_string
##  Length:9447            Length:9447               
##  Class :character       Class :character          
##  Mode  :character       Mode  :character          
##                                                   
##                                                   
##                                                   
##

The Descriptive statistics indicates that

  • All the courses in the data set were paid for and non was in the wish list
  • Number of Subscribers per course ranged from 3 to 374836 with an average number of subscribers as 3399, this is a huge variance, which will be investigate into.
  • Number of published lectures ranged from 0 to 699 with an average of 34 lectures per course.
  • 510 courses had discounts applied to them
  • The prices also ranged from 1280 Indian Ruppies (INR) to 12800 INR, but the average course was 4829 INR.
  • The courses were created between 14th April, 2010 and 2nd September, 2020.

Overall, the course has a lot of variance, that is some courses were extremely more expensive than others and some performed extremely higher than most of the courses. Outlier courses were many

4.2 Data Distribution

This will give an overview of how the data is distributed, how far or near they are from the center.

4.2.1 Distribution of Number of Subscribers
#ggplot will be used for visualization

ggplot(data= courses, mapping = aes(x= num_subscribers)) + geom_histogram(fill="blue") +
    geom_freqpoly(color="red") +
    labs(title= "Distribution of Number of Subscribers", x="Number of Subcribers", y= "Number of Courses")
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

  • Number of Subscribers is skewed to the left with a long tail on the right side indicating that some courses had extremely more number of subscribers than others.
4.2.2 Distribution of Number of Average rating
#ggplot will be used for visualization

ggplot(data= courses, mapping = aes(x= avg_rating)) + geom_histogram( fill= "blue") +
    geom_freqpoly(color="red") +
    labs(title= "Distribution of Number of Average ratings", x="Average Rating", y= "Number of Courses")
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

  • Average rating of the courses were mostly above 3.3 towards 5.0, indicating that they most of them performed above average, however there were a few courses that performed very poorly and fell between 0.5 and 3.0
4.2.3 Distribution of Number of Reviews
#ggplot will be used for visualisation

ggplot(data= courses, mapping = aes(x= num_reviews)) + geom_histogram(fill="blue") +
    geom_freqpoly(color="red") +
    labs(title= "Distribution of Number of Reviews", x="Number of Reviews", y= "Number of Courses")
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

  • The Distribution of number of reviews looks similar to that of the number of subscribers, which is normal because as a course gets more subscribers, it is expected to get more reviews for it.
4.2.4 Distribution of Discounted courses

To get the number of discounted courses, we will use the mutate function to create a column called is_duplicated, this contain a boolean indicating where or not a course has a discount

# All discounted courses have   `NA` at the column `discount_price__amount`
courses <- courses %>% 
  mutate(is_discounted = ifelse(is.na(discount_price__amount)==FALSE, TRUE, FALSE ))
# Creating a Bar Chart to show distribution of discounts
ggplot(data=courses, mapping = aes(x= is_discounted, fill= is_discounted)) +  geom_bar() +
    labs(title= "Distribution of Discounts", x="Course Discount Status", y= "Number of Courses")

  • Majority of the Courses in have discount prices attached to them
4.2.5 Distribution of Discounted prices

I will first filter out the discounted courses

discounted_courses <- courses %>% filter(is_discounted==TRUE)
# Creating a Histogram
ggplot(data= discounted_courses, mapping= aes(x= discount_price__amount)) + geom_histogram(fill="blue") +
    labs(title= "Distribution of Discounted amounts", x="Amount of Discount", y= "Number of Courses")
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

  • This does not give a clearer picture of the discount distribution since the prices of the courses differ, so I will create a column to calculate the discount percentage
discounted_courses <- discounted_courses  %>% 
  mutate(discount_percentage= (discount_price__amount/ price_detail__amount)*100) 
# Creating a Histogram
ggplot(data= discounted_courses, mapping= aes(x= discount_percentage)) + geom_histogram(bins= 10, fill="blue") +
    labs(title= "Distribution of Discount Percentages", x="Discount Percentage", y= "Number of Courses")

  • Majority of the discounts were between 4% and 35% with a few above 40%
4.2.6 Trend of number of created courses Since 2010

We will first create a column for the year of course upload separately

courses <- courses %>% mutate(year_published=year(published_time))
# Creating a Bar Chart for each year
ggplot(data= courses, mapping= aes(x= year_published, fill=is_discounted)) + geom_bar()

  • There was a rise in course publishing from 2010 to 2018, except for a flatuation in 2016. +More data is needed to prove that total number of courses created started to decline after 2018, since we do not have complete data for 2020 and 2018 could have been a very high spike.

4.3 Correlation Analysis

Correlation Analysis check how various Variables relate to each other. This analysis will be done by calculating cor and also drawing out a scatter plot to indicate data relations while differentiating between discounted courses and non-discounted courses

4.3.1 Correlation between number of subscribers and number of reviews

Does More Subscribers mean More reviews?

#calculating correlation
correlation <- courses %>% select(num_subscribers, num_reviews) %>% cor()
correlation<- round(correlation[1,2],3)

ggplot(data= courses, mapping= aes(x= num_subscribers, y= num_reviews, color= is_discounted)) + geom_point() + 
    labs(title= "Number of Reviews vs Number of Subcribers", x="Number of Subcribers", y= "Number of Reviews") + 
  annotate("text", x= 250000, y= 30000, label= paste("Correlation= ",correlation))

  • There is a strong relationship between number of subscribers and number of reviews.
  • More Subscribers Corresponds to more Reviews
4.3.2 Correlation between number of published lectures and number of subscibers

Does more lectures correspond to more Subscribers?

#calculating correlation
correlation <- courses %>% select(num_subscribers, num_published_lectures) %>% cor()
correlation<- round(correlation[1,2],3)

ggplot(data= courses, mapping= aes(x= num_published_lectures, y= num_subscribers, color= is_discounted)) + geom_point() + 
    labs(title= "Number of Publised Lectures vs Number of Subcribers", x="Number of Published Lectures", y= "Number of Subscribers") + 
  annotate("text", x= 450, y= 200000, label= paste("Correlation= ",correlation))

  • There is a weak relationship between number of published lectures and number of subscribers
  • More Lectures does not mean more Subscribers
4.3.3 Correlation between Average Review and number of Subscribers

Does Better Reviews Mean More Subscribers?

#calculating correlation
correlation <- courses %>% select(num_subscribers, avg_rating) %>% cor()
correlation<- round(correlation[1,2],3)

ggplot(data= courses, mapping= aes(x= avg_rating, y= num_subscribers, color= is_discounted)) + geom_point() + 
    labs(title= "Average Rating vs Number of Subcribers", x="Average Rating", y= "Number of Subscribers") +
  annotate("text", x= 3.5, y= 100000, label= paste("Correlation= ",correlation))

  • There is a very weak relationship between Average rating and Number of Subscribers
  • However, Poor reviews lead to very poor performance
  • Better Review Does not Always mean More Subscribers
4.3.2 Correlation between course price and number of Subscribers
#calculating correlation
correlation <- courses %>% select(num_subscribers, price_detail__amount) %>% cor()
correlation<- round(correlation[1,2],3)

ggplot(data= courses, mapping= aes(x= price_detail__amount, y= num_subscribers, color= is_discounted)) + geom_point() +
    labs(title= "Course Price vs Number of Subcribers", x="Course Price", y= "Number of Subscribers") + 
  annotate("text", x= 5000, y= 200000, label= paste("Correlation= ",correlation))

  • There is a weak relationship between Course Price and number of Subscribers
  • Price Matter a little, when it comes to getting more subscribers
4.3.5 Correlation between period of publication and number of Subscribers
# We will create a column to calculate the number of days since course was published with today as reference
courses <- courses %>% mutate(days_since_published = as.numeric(now() - published_time))
#calculating correlation
correlation <- courses %>% select(num_subscribers,days_since_published) %>% cor()
correlation<- round(correlation[1,2],3)

ggplot(data= courses, mapping= aes(x= days_since_published, y= num_subscribers, color= is_discounted)) + geom_point() + 
    labs(title= "Number of Days Since Published vs Number of Subcribers", x="Number of days since Published", y= "Number of Subscribers") + annotate("text", x= 3500, y= 200000, label= paste("Correlation= ",correlation))

  • Number of days of course been on the Udemy Platform has a weak relationship with number of subscribers
  • The Longer the course on the platform does not translate to more subscribers
4.3.6 Correlation between discount and number of Subscribers
#calculating correlation
#correlation <- courses %>% select(num_subscribers, is_discounted) %>% cor()
#correlation<- round(correlation[1,2],3)

ggplot(data= courses, mapping= aes(x= is_discounted, y= num_subscribers, color= is_discounted)) + geom_point() + 
    labs(title= "Discount Status vs Number of Subcribers", x="Discount", y= "Number of Subscribers")

  • Courses with Discount tend to perform better than those without discount. Discount is important in getting more Subscribers
4.3.7 Correlation between Discount percentage and number of Subscribers

I will first assign 0 to courses without discount

courses <- courses %>% mutate(course_discount = ifelse(is_discounted==FALSE, 0, discount_price__amount))

# Creating a Discount percentage column
courses <- courses %>% mutate(discount_percentage = (course_discount/ price_detail__amount)*100)

#calculating correlation
correlation <- courses %>% select(num_subscribers, discount_percentage) %>% cor()
correlation<- round(correlation[1,2],3)

ggplot(data= courses, mapping= aes(x= discount_percentage, y= num_subscribers, color= is_discounted)) + geom_point() + 
    labs(title= "Discount Percentage vs Number of Subcribers", x="Discount Percentage", y= "Number of Subscribers") +
  annotate("text", x= 30, y= 200000, label= paste("Correlation= ",correlation))

  • There is weak negative relationship between discount percentage and number of subscribers
  • More Discount does mean more Subscriber, it can mean less

5. Share Phase

A PowerPoint Presentation will be created with the key findings from this data. But below are the key findings of the Analysis.

5.1 Findings

  • Some courses perform extremely better than the rest, making the data skew to the left with a very long tail on the right.
  • High Number of Subscribers corresponds to high number of reviews and the vice versa. More reviews can lead to more Subscribers.
  • Number of lectures, Average Rating, Course Price, Number of Days since course was created, and Discount percentage has little effect on the number of subscribers.
  • Poor reviews leads to very poor performance.
  • However Discounted courses tend to perform better than Non-Discounted Courses.

5.2 Limitations

These were the limitations encounted in the analysis.

  • I have yet to learn Text Analysis, which will help me to analyse how course titles and certain keywords influence Course performance.
  • The data did not include the specific category of course to allow for Segment Analysis.
  • The Outliers in the dataset were many and may have greatly influenced the outcome.
  • Description of Thumbnails were not included for the analysis.

5.3 Conclusion

  • Having a discount on course and more Reviews can positively improve the performance of a course on UDEMY.
  • Other parameters like course price, average rating, how long the course stays on the platform and number of lectures on a course has little effect on it’s performance

6. Act Phase

After the insights comes the action. Knowledge without action is useless. From the Analysis, I suggest these actions to each stakeholder

6.1 Recommendations to Course Creating Manager

  • Create courses that are more engaging than courses that are too long.
  • Continually remind Subscribers to write reviews on the course, as this will greatly influence the performance of the course.

6.2 Reccomendations to Marketing Director

  • Incorporate Discounts more often into courses, as this can help the course to attract more subscribers.
  • Very High course discount may reduce the credibility of the course and reduce the number of subscribers
  • Cathy Titles and thumbnails should be used, but it is important to note that we are yet to gather more data to analyse how each affect course performance.

6.3 Reccomendation to Chief Data Analyst

  • Text Analysis should be carried out on the course titles to know which keywords influence course performance.
  • More Data should encompassing Thumbnail, Course Category, and other performance influencing variables should be gathered for further analysis.

6.3 Reccomendations for Further analysis

  • Do Text Analysis
  • Outliers should be removed analysed separately from the rest for better results.
  • A more detailed dataset should be used.